Multi-rotor UAS for precision crop-spraying: developing an adaptive guidance algorithm to minimize spray drift in a wind environment

In recent years, modern precision agriculture techniques led to a rise in the use of Unmanned Aerial Vehicles (UAVs) for many different purposes, such as land survey, plant health monitoring and crop spraying. These vehicles are particularly suited for the use over rough and mountainous terrains or dispersed plots of land, that traditional land vehicles may find difficult to reach or too time consuming. UAVs can also benefit the farmers as they can be, and often are, automated in their operation, further increasing productivity while reducing the exposure to potentially hazardous chemicals. Politecnico di Torino, in collaboration with SEASTAR, started the research on "New technical and operative solutions for the use of drones in Agriculture 4.0" through a grant linked to the PRIN 2017 (\textit{Progetti di ricerca di Rilevanza Nazionale}, Research projects of national interest), a program of the Italian Ministry of Education, University and Research. This led to the preliminary design of a drone for crop spraying on vineyards, as this cultivation is characterised by uneven, sloped terrain, it presents the ideal scenario to evaluate the precision of Plant Protection Products (PPP) deposition and is widely cultivated throughout Italy. The thesis here presented is the natural continuation of the previous work. The aim was to develop a guidance and control algorithm for a multi-copter that would be able to minimize PPP particles drift, taking into account the wind conditions. In order to do so, a particle drift model was developed as well, based on an experimental campaign in a wind tunnel. Chapter 1 introduces the work, reviewing the state of the art for this application. In Chapter 2, the methodology for the development of the drift model is presented. Then, in chapter 3 the multi-rotor platform, its working principles and dynamics are discussed. The guidance and control system and its integration with the drift model can be found in chapters 4 and 5. All results are then presented and discussed in chapter 6, while chapter 7 draws the conclusions and explores possible future developments.